Download 07.Discussion

4. Discussion
69
4. DISCUSSION
The research done here proved that large-scale whole mount in situ
hybridization is an efficient way to screen genes involved in embryogenesis
from cDNA libraries. The advantage of the approach is that it allows direct
identification of single clones and hence single genes according to expression
patterns. 29 of 384 single clones were sorted out because of specific expressions.
Subsequently, 3 of 29 genes, namely, XODC2, XCL-2, and XETOR were
selected for further study based on comparison with Genbank databases and
their expression patterns. Actually, XODC2 is a piece of evidence that there
exists a second type of ornithine decarboxylase in specific cells besides the
ubiquitous form, although no solid indication was acquired in the present study
that it plays a significant role in embryonic development. While the other two
genes, XCL-2 and XETOR, were shown to play pivotal functions respectively in
two developmental procedures: morphogenetic movements and primary
neurogenesis.
4.1 XCL-2 and its role during embryogenesis
4.1.1XCL-2 is a novel m-type Calpain and disrupts morphogenetic
movements during embryogenesis in Xenopus laevis
It has been shown that XCL-2 encodes a large subunit of m-type Calpain, a
novel family member of the calcium-dependent proteases. Typical Calpains,
either ubiquitous or tissue-specific, are around 700 amino acids in length and
consist of four domains. The first domain is responsible for autolysis, which
reduces calcium requirement for proteolytic activity; the second domain has
three active sites, Cys 105, His 262 and Asn 286, for proteolytic activity; the
third domain has an unknown function; and the fourth domain, comprising EFhand motifs, is for Ca2+ binding. This domain structure and the active sites are
4. Discussion
70
conserved in all typical Calpains identified so far, including XCL-2 in the
present study. Therefore, it is reasonable to assume that XCL-2 also possesses
the calcium-dependent proteolytic activity. As a prerequisite for functional
analyses, whole mount in situ hybridization and RT-PCR were performed to
examine the expression patterns of XCL-2 during X. laevis development. Using
whole mount in situ hybridization, embryos showed no detectable signals earlier
than the late gastrula stage. Expression was first detected in the area close
beneath ventral blastopore lip at around stage 12.5. During neurulation, signals
were found in the mesoderm-free zone at the most anterior and ventral parts of
the circumblastoporal collar at the most posterior zone. At the tailbud stages,
expression was restricted to the cement gland and proctodeum, and to the
cement gland only at the late tailbud stages. Using RT-PCR, weak expression
could already be observed at stage 10. It also demonstrated the tissue-specific
expression of XCL-2 in adult tissues, such as brain, eye, heart, intestine, kidney,
lung, stomach and testis. The relatively high level of expression in stomach
suggests that XCL-2 and nCL-2 are evolutionarily conserved homologues. The
latter is predominantly expressed in rat stomach.
The restricted expression of XCL-2 in late gastrulae or early neurulae in the
ventral circumblastoporal collar could be suggestive for its function. The
circumblastoporal collar, including the dorsal and ventral parts, represents a
massive accumulation of mesodermal cells. From this region, dorsal axial
mesoderm is continuously generated by radial intercalation and convergent
extension. This was the initial impetus to investigate whether XCL-2 could be
involved in cell movements during X. laevis early embryogenesis.
Overexpression of XCL-2 and a dominant-negative mutant, C105S, were
therefore carried out to investigate its functions during early embryogenesis.
Injections were made into either two dorsal blastomeres or two ventral
blastomeres at the 4-cell stage. It was found that overexpression at the vegetal
pole did not affect embryonic development significantly, in contrast to
overexpression at the animal pole or even the equatorial region. This is probably
due to the fact that the injected RNA at the vegetal pole will mainly be
distributed in the endoderm and therefore it cannot exert its activity in the
circumblastoporal collar. Therefore the effect of overexpression of the mRNAs
at the animal pole was studied.
4. Discussion
71
Overexpression of XCL-2 results in a delay of the involution of mesoderm.
At the beginning of gastrulation, the blastopore in injected embryos formed
normally, as in uninjected control embryos. However, during gastrulation the
blastopore of injected embryos failed to close and a large yolk plug was
observed in subsequent developmental stages. This disturbance of involution
generates a significant phenotype. The results suggest that overexpression of
XCL-2 causes the disruption of the blastopore and disturbance of the gastrulation
movements. It was examined whether there are obvious cell fate changes in
these affected embryos using mesodermal or neural markers, including Xbra,
Chordin, Xvent-1, XMyoD, Xotx2 and Xsox3. In fact, all of these markers are
expressed at similar levels to the controls but are not correctly localized, which
suggests that no obvious cell fate alteration takes place, but rather distinct
changes in cell migration. Furthermore, it was shown with Chordin, XMyoD and
Xsox3 in injected embryos that during gastrulation and neurulation the
mesoderm and the posterior neural plate do not converge towards the midline, as
in normal embryos. These data suggest that XCL-2 participates in the
convergent extension movements starting from the midgastrula stage.
It was further revealed, by overexpressing a dominant-negative-type mutant
C105S, that XCL-2 activity is required for morphogenetic movements.
Overexpression of C105S and wild-type XCL-2 resulted in similar phenotypes
(open blastopore and bifurcate tail). These on the first glimpse confusing results
could be explained by the observation that both overexpression of wild-type
XCL-2 or inhibition (partial or total loss of function) by a dominant-negative
mutant, C105S, will cause the disturbance of morphogenetic movements. These
data are in agreement with results using other genes, where overexpression of
both wild-type and truncated forms of Wnt11 (Tada and Smith, 2000) or
Frizzled-7 (Djiane et al., 2000) also blocks convergent extension movements
and generates similar phenotypes. However, our histological sections show that
the embryos injected with C105S form distinct bifurcate notochord
malformations, in contrast to wild-type XCL-2 overexpression. This phenotype
can be rescued by coinjection of XCL-2 and C105S mRNA at the proper ratio.
These results suggest that the mutant acts specifically in vivo by competition
with wild-type protein for proteolytic substrates. However, because Calpains
have been identified as a large gene family in other vertebrates, such as human
4. Discussion
72
and rat, it can be suggested that such a family also exists in X. laevis. Therefore,
the possibility cannot be ruled out that the dominant-negative mutant C105S also
suppresses other related members. In summary, present data suggest that XCL-2
is a prerequisite for morphogenetic movements during early embryogenesis in X.
laevis. It is highly probable that Calpain regulates cell movements by changing
cell adhesive activity via proteolytic cleavage of proteins that are essential for
morphogenesis, but this remains to be elucidated.
4.2 XETOR and its role during primary neurogenesis
Here in this study it is presented for the first time the expression and
function of an ETO related gene, XETOR, during primary neurogenesis in
Xenopus laevis. Like other members of ETO oncogene family, XETOR is also
expressed primarily in the nervous system. Both gain- and loss-of-function
studies showed that XETOR plays key roles in primary neurogenesis. It inhibits
primary neuron formation by establishing a negative feedback loop with
proneural genes. This inhibition is not mediated by lateral inhibition signaling
but a result of an independent action. Lateral inhibition and XETOR antagonize
each other but both are required for primary neurogenesis. They consist of a dual
inhibitory mechanism to refine the exact localization and number of primary
neurons.
4.2.1 XETOR is an inhibitory factor for primary neurogenesis in
independence of lateral inhibition
XETOR encodes a putative protein that shares all characteristics of other
members of oncoprotein family ETO/MTG8: the four conserved Nervy
Homologous Regions and the two unusual zinc-finger motifs. Besides the
common structure, another feature shared among these genes examined,
including Nervy in Drosophila, is that they all have expression in the nervous
system (Feinstein et al., 1995; Wolford and Prochazka, 1998). This suggests a
conserved function of ETO proteins in nervous system. In Xenopus laevis,
expression of XETOR during primary neurogenesis begins at stage 12.5 in a
pattern of longitudinal stripes at either side of dorsal midline, similar to the
patterns of primary neuron marker genes. Double in situ hybridization showed
that expression domain of XETOR is overlapping with, however broader than
4. Discussion
73
that of N-tubulin, confirming that XETOR is another marker gene for primary
neurogenesis. The broader expression domain of XETOR may suggest that it
plays to refine the localization of primary neurons, and this has been confirmed
by the study. Temporally, expression of XETOR starts at stage 12.5, following
that of Xngnr-1, Xash-3, XMyT1, X-Delta-1, Xath3 and preceding that of
XNeuroD, but at the same time when the differentiated neuron marker gene Ntubulin is turned on. This temporal sequence should suggest that XETOR is
turned on when neuronal differentiation begins and hence a role in regulating
neuronal differentiation.
Further lines of evidence support this idea more directly. It was proved that,
on the contrary to promoting neuronal differentiation as bHLH proneural
proteins do, it acts to repress the procedure. First, gain-of-function data show
that XETOR inhibits expression of neuron marker genes N-tubulin and Xaml.
This loss of primary neuron formation is not a result of disruption of neural
plate, as indicated by intact expression of Xsox3 in response to XETOR
ovexpression. Second, overexpressed XETOR extirpates the neuron-inducing
activity of proneural genes, Xash-3, Xath3 and XNeuroD, as well as the zincfinger gene XMyT1, as revealed by coinjections of XETOR together with any
one of these genes. Therefore, XETOR should be able to inhibit primary neuron
formation via repressing the function of proneural genes. Third, inhibition of
XETOR function in vivo leading to a neurogenic phenotype of expanded
neurogenic domain also suggests that XETOR is a key negative regulatory
factor for primary neurogenesis.
It seems that a negative feedback loop is established between XETOR and
proneural genes. The reason is that expression of XETOR is exclusively
activated or promoted by the genes examined in the study except the zinc-finger
gene XMyT1, and in turn, XETOR inhibits the function of these genes, with the
exception of Xngnr-1. Although all proneural genes can activate XETOR
expression, it is proposed here that XETOR is more likely to be activated directly
by Xath3 than by Xngnr-1. The first reason is that Xngnr-1 activates Xath3 in a
unidirectional way and Xath3 is a direct downstream target of Xngnr-1. The
second reason is that, in temporal sequence, XETOR begins to express at stage
12.5 almost immediately after Xath3, which starts to express at stage 12 (Perron
et al., 1999). Therefore, the temporal expression of XETOR suggests that it
4. Discussion
74
functions in the late stage of primary neurogenesis. This is also confirmed by the
fact that XETOR does not affect the expression or/and function of neuronal
determination genes Xngnr-1 and Xash-3, but instead inhibits the function of
neuronal differentiation genes.
In the data and discussion above, it can be noticed that XETOR and lateral
inhibition signaling have certain characteristics in common: both are activated
by proneural genes, and conversely, both inhibit the function of proneural genes
and hence primary neuron formation. Considering the similarity between the
expression of XETOR, X-Delta-1 and other components in lateral inhibition, it
was asked whether XETOR is also a component of this signaling pathway such
that XETOR function is orchestrated by lateral inhibition. A few lines of
evidence negate the idea but support that XETOR and lateral inhibition are two
different working mechanisms. The first point is that expression of X-Delta-1 is
repressed in response to overexpressed XETOR, and vice versa, XETOR
expression is inhibited by activated lateral inhibition signaling pathway. In
addition, when lateral inhibition is blocked, XETOR expression is promoted.
This phenomenon is just contrary to that for Notch targets, which are promoted
by activated lateral inhibition pathway while repressed by the blocked pathway.
Therefore, lateral inhibition and XETOR is a pair of antagonists. The second is
that XETOR still efficiently inhibits N-tubulin expression in the absence of
lateral inhibition, as shown by coinjection of XETOR and X-Delta-1STU. The
third is that the expression of Notch targets, ESR1 and XNAP, is not affected
either with XETOR or without XETOR. Finally, that Xngnr-1 is refractory
while XNeuroD is sensitive to XETOR activity suggests again that XETOR
should not be a component of lateral inhibition pathway, because it was
previously shown that early proneural genes as Xngnr-1 and Xash-3 are sensitive
to lateral inhibition while late genes as XNeuroD and Xebf3 are refractory.
4.2.2 The molecular mechanism for transcriptional repression
activity of XETOR
The initial idea for constructing truncation mutants of XETOR was to test
whether any one of them could function antimorphically to the wild type
protein, such that loss-of-function assays could be performed. Unexpectedly, all
these truncation mutants exclusively exhibit a repression effect, showing that
4. Discussion
75
XETOR is a robust transcription repressor. The data on these mutants revealed
important information on the function of different Nervy Homologous Regions
(NHRs). One principal mechanism for transcriptional repression is the
modification of chromatin conformation by histone deacetylases (HDACs; for
reviews, see Grunstein, 1997; Struhl, 1998; Torchia et al., 1998). ETO has been
identified as a potent repressor of transcription by recruitment of HDAC1 and 2
via corepressors N-CoR (nuclear receptor corepressor), mSin3 or SMRT
(silencing mediator for retinoid and thyroid-hormone receptors) (Gelmetti et al.,
1998; Lutterbach et al., 1998; Wang et al., 1998). Moreover, ETO itself acts also
as a corepressor recruited by promyelocytic leukemia zinc finger protein
(PLZF), which mediates transcriptional repression via the action of HDACs
(Melnick et al., 2000). Multiple regions in ETO have been shown to cooperate in
transcriptional repression. In the four conserved NHRs, both NHR3 and NHR4
but not NHR4 alone were shown to be essential for the interaction between ETO
and N-CoR (Hildebrand et al., 2001). However, the binding of N-CoR with
NHR3 and 4 per se is not sufficient for the mediation of transcriptional
repression. While the core repressor domain (CRD), consisting of NHR2 and its
N- and C-terminal flanking sequences, is the smallest region that shows
significant repression activity on its own. Furthermore, this domain interacts
strongly with mSin3A (Hildebrand et al., 2001). It is somewhat mystic for the
function of NHR1, as ETO short of this region still induces maximal repression.
While it is interesting for the sequence between NHR2 and NHR3, because it
was clarified as the binding site for PLZF, and ETO short of this part
significantly decreases repression activity (Melnick et al., 2000; Hildebrand et
al., 2001).
The information on ETO domain functions may help explain the results of
overexpression of XETOR truncation mutants. It was observed that the
inhibitory effects of overexpression of truncation mutants p13#trunc1, 2, 4, and
5 on primary neuron formation are indistinguishable from that of the wild type
XETOR. Therefore, these data suggest that single truncations of either NHR1 or
NHR4, or double truncations of either NHR1 and 2 or NHR3 and 4 have no
dramatic influences on their repression activity. The data are reasonable because
the CRD homologous sequence is intact in p13#trunc1, 2 and 4. Especially,
p13#trunc4 should retain the maximal repression activity because NHR1 is not
4. Discussion
76
essential for ETO activity. Although p13#trunc3, containing only the CRD
homologous sequence, showed also repression activity, it functions not as
efficiently as other mutants. These data suggest that CDR need the cooperation
with either NHR1 (as in p13#trunc2) or NHR3 and 4 (as in p13#trunc4) to exert
a similar level of transcriptional repression activity to that of the wild type
protein. However, based on the information available so far, it is not readily able
to explain the result of p13#trunc5 overexpression, because this mutant showed
similar inhibitory effect to those of other mutants. In the case of ETO as
mentioned above, NHR3 and 4 can bind N-CoR, but this binding by itself is not
sufficient for the mediation of transcriptional repression. The present data
suggest that XETOR may mediate transcriptional repression not only through
machineries that are well known, but also those that are not well known. Up to
today, XETOR has been the only protein of ETO oncoprotein family identified
in Xenopus. Therefore, it is difficult to conclude whether it is an orthologue or
paralogue of ETO. According to amino acid identity data, XETOR is more
closely related to MTGR1 (72%) than to ETO (59%), hence it is more likely to
be a paralogue. There is evidence that MTGR1 tends to form heterodimers with
ETO via the oligomerization domain NHR2. It remains to be elucidated that if
this also holds true in Xenopus.
4.2.3 XETOR and lateral inhibition comprise a dual inhibitory
mechanism to refine the number and localization of primary
neurons
The data have shown that XETOR can be exclusively activated by proneural
genes. Conversely, XETOR tends to inhibit the neuron-inducing functions of
proneural genes, with the exception of Xngnr-1. Thus a negative feedback loop
is formed between XETOR and proneural genes. Such a responding mechanism
is reminiscent of that between proneural genes and lateral inhibition signaling.
Are these two the same machinery? There is evidence that they are different
because they inhibit each other and work independently, as discussed above.
Why do they coexist during certain period of primary neurogenesis?
Loss-of-function assay discloses important clues for such a question. When
XETOR alone is knocked out, a neurogenic phenotype of significantly enlarged
neurogenic domain is resulted. It is in congruence with the fact that the
4. Discussion
77
expression domain of XETOR is broader than, though overlapping with, that of
N-tubulin. However, the density of neurons does not vary appreciably. These
data suggest a role of XETOR in refining the localization of primary neuron
formation. The nearly normal salt-and-pepper pattern of N-tubulin expression
should be generated by an increased level of lateral inhibition signaling, because
XETOR can antagonize lateral inhibition by disabling the ligand gene X-Delta1, and consequently an enhancement of lateral inhibition could be expected
when XETOR activity is repressed. It was confirmed by double depletion of
XETOR and lateral inhibition, which caused a phenotype of increased density of
neurons in enlarged proneural domain. Such a phenotype is apparently distinct
from that generated either by depletion of XETOR alone (enlarged proneural
domain with an unaltered density of neurons) or by depletion of lateral
inhibition alone (increased density of neurons with an unaltered proneural
domain). These data suggest that XETOR and lateral inhibition are both required
for primary neurogenesis: lack of either one will lead to overproduction of
primary neurons either in an extra large sized neurogenic domain or in an
excessive density. Hence, lateral inhibition and XETOR must cooperate to
define the exact number and localization of primary neurons.
The unidirectional genetic cascade Xngnr-1-Xath3-XNeuroD underlies the
major pathway promoting neurogenesis, which is in turn regulated by lateral
inhibition. At the beginning of primary neurogenesis, expression of Xngnr-1
defines the proneural domains where primary neurogenesis will occur. As a
result of lateral inhibition, Xngnr-1 makes neuroectodermal cells competent to a
neuronal state. While this state is unstable, genes like XMyT1, Xath3 and XCoe2
are subsequently activated by Xngnr-1 to stabilize the competent state
(Bellefroid et al., 1996; Takabayashi et al., 1997; Dubois et al., 1998; Perron et
al., 1999). By this way early proneural genes and lateral inhibition establish a
mechanism for defining the number of primary neurons. However, this is
possibly only part of the landscape for primary neurogenesis. It is known that
proneural gene Xngnr-1 defines a proneural domain much broader than the area
where primary neurons form. Similarly, the expression domains of early
expressed genes are much broader than that of late expressed proneural genes.
Such a correlation of temporal and spatial expression is also reported between
XMyT1 and N-tubulin (Bellefroid et al., 1996). Therefore a regulatory system
4. Discussion
78
should be there for the refinement of the neurogenic domains in the later period
of primary neurogenesis. Proneural genes should be not the candidates because
the consistently promote neuron formation. Neither is lateral inhibition because
it is primarily responsible for cell fate selection, and knockout of lateral
inhibition does not alter the proneural domain size but only the density of
primary neurons. It is proposed here that XETOR is such a factor because of a
few lines of evidence. First, the most direct evidence is that eradication of
XETOR function in vivo results in an enlargement of proneural domain, as
discussed above. Second, XETOR begins to express at a time later than X-Delta1 and does not inhibit neuron determination genes, especially Xngnr-1. This
feature of XETOR function will ensure the initiation of early neuronal
determination, which is regulated by lateral inhibition. Third, evidence has
shown that each proneural gene activates three things: X-Delta-1, XETOR and
the downstream proneural genes. Considering additionally that X-Delta-1 is
activated earlier than XETOR and that X-Delta-1 and XETOR antagonize each
other, it would be reasonable to assume that some cells expressing X-Delta-1
will not express XETOR, and vice versa. Due to the inhibitory effect of XETOR
on the expression and function of proneural genes, the neurogenic domain
defined by the downstream proneural gene will be more restricted than the
domain defined by the upstream gene.
Although late expressed neuronal differentiation genes, as XNeuroD and
Xebf3, are resistant to lateral inhibition, they still activate lateral inhibition. This
activation might be reasonable because it can serve to regulate the expression of
XETOR during this stage of primary neurogenesis.
Based on the data and discussion above, a fresh model for primary
neurogenesis is summarized as follow: expression of proneural gene Xngnr-1
marks the initiation of primary neurogenesis and defines the proneural domain.
At the same time, Xngnr-1 also activates downstream proneural genes and
lateral inhibition. Some cells are committed to a neuronal fate and others remain
ectodermal via a negative feedback loop. In this way lateral inhibition
determines the density of primary neurons. However, the size of neurogenic
domain where primary neurons will form is not determined. At the time
approximately when neuronal differentiation begins, XETOR is activated also by
proneural genes. Because XETOR expression is antagonized by lateral
4. Discussion
79
inhibition, and XETOR represses the expression and function of late expressed
proneural genes, thus it is possible to restrict the neurogenic domain to a correct
localization.
The functions of proneural and neurogenic genes are all proved to be
conserved throughout the spectrum of organisms from Drosophila to Xenopus. It
has been shown that there also exist XETOR homologues in other organisms,
including Nervy in Drosophila. Moreover, the expression of these homologues
is restricted to the nervous system, too. It is logic to deduce that certain XETOR
homologue in other organisms play the same or at least similar role. Hence the
dual inhibitory mechanism identified in this study may also conserve in other
organisms.
It should be mentioned in passing that XETOR could also be involved in
hematopoiesis. The XETOR homologue ETO/MTG8 in human is often
translocated to AML1 to make a fusion transcript in acute myeloid leukemias. It
has been shown that Xaml, the AML1 homologue in Xenopus laevis, functions
in the specification of hematopoietic stem cells in vertebrate embryos (Tracey et
al., 1998). Considering their expression in both primary neurons and the
presumptive ventral blood island, therefore it is proposed here that XETOR may
also play a role in primitive hematopoiesis.
In summary, based on the present data, it is proposed here that during
primary neurogenesis, lateral inhibition and XETOR comprise a dual inhibitory
mechanism to refine the exact number and localization of primary neurons, via
repression of the expression and function of proneural genes.